The purpose of this article was to examine the performance of Shewhart average and standard deviation control charts in relation to the probability of type I error at the occurrence of normality assumption violation. A simulation of a series of 500,000 samples (subgroups) of sizes n=3, 5, 7, 10, 15, 20 and 25 was performed. The samples were simulated from the following distributions: Normal, Binomial, Poisson, Exponential, Chi-square, Gamma and Weibull. It was possible to find that, in non-normal data, there was an increase in the probability of type I error on the average chart of all simulated distributions. The minimum sample size required is related to the degree of asymmetry of the data distribution and, in some cases, not even n=25 presented satisfactory results. In the S chart, the increased probability of type I error is significantly higher for almost all distributions simulated and its performance is influenced not only by the type of distribution but also by the sample size.
Shewhart control charts; Normality assumption; Type I error; Robustness
